I've been experimenting with a 1% thermistor from US Sensor, manuf. part# GP104L8F. This is a 1% 100K thermistor rated up to 300C that I've been experimenting with on our PrismX with the J-Head MKV-B. The GP104L8F is pricey but I've had my eye on it and my Pronterface temperature graph's fluctuations for a while, so decided to give it a shot on our new machine.
It's only been used for ~200 printing-hours so far but I can report some behaviors. This is anecdotal but so far it seems the temp stays within +/- 2C of the target temperature consistently, as would be expected at 200C. The visual temperature graph ala Pronterface is certainly noticeably flatter than any of the other 3% (or 10% ouch) therms I've experimented with.
To generate a thermistor table I used the Sparkfun Oven Controller - with it's MAX6575 ADC thermocouple. Unfortunately, I don't know how the ADC on the oven controller was calibrated, but this is the best reference we have available in our shop currently. During measurements the thermocouple probe was placed physically near the thermistor. I also had to modify Marlin to respond to a custom GCode command with the ADC reading from the hotend.
I used M777 to return the ADC value from extruder 0 in Marlin.CPP. Here's the snippet:
[
gist.github.com]
Finally, I used code at [
hydraraptor.blogspot.com], which was merged into Marlin for the latest update of createTemperatureLookupMarlin. Only the original update provided the ADC-based functionality. Marlin's only uses resistance readings. It seems to me Marlin should bring back the ADC support in the original code and add M777 (or other number) support - so users with better references (than my ADC) can re-calibrate. In general this use-case seems valuable.
For anybody who's interested I will post the therm table to Marlin and can follow up with a link. Right now...the low-end of the table, near room-temperature, is inaccurate and I want to improve that before posting.
Edited 1 time(s). Last edit at 05/20/2013 06:33PM by tenacious.
